---
title: "From Collateral to Cash Flow: The 'Secondary Licensing' Model That Would Make Data-Asset ABS Real"
author: "DCC Editorial"
published: 2026-06-05T02:00:00.000Z
url: https://datacompliancechina.com/posts/data-asset-abs-secondary-licensing-cash-flow/
description: "If today's data-asset ABS is '1.0' — data as collateral behind a conventional debt claim — then '2.0' is the version where the data's own cash flow (licensing fees, data-service subscriptions) directly repays the securities, upgrading data from credit-enhancement tool to genuine underlying asset. This third brief in DCC's data-asset-ABS series examines the structure most likely to get there: the 'secondary licensing' (二次许可) model borrowed from intellectual-property ABS, in which a holder exclusively licenses data to an originator for an upfront lump sum, then takes a reverse exclusive licence back and pays periodic fees that become the ABS cash flow — ownership never moving. It maps the obstacles (data's non-exclusivity defeats 'exclusive licence' and 'exclusive possession'; PIPL/DSL cap what can be licensed; valuation is immature), the finance-lease-of-data variant, and the early policy encouragement (Anhui's March 2026 measures endorsing reverse-licensing). The irony the June 2026 halt exposed: regulators want real data cash flow — which is exactly what 2.0 promises but cannot yet deliver at scale."
tags: ["data-economy", "data-asset-abs", "securitisation", "secondary-licensing", "data-property-rights", "data-licensing", "finance-lease", "financing"]
laws_cited: ["data-foundation-system-opinions", "data-property-rights-registration-guide-draft", "pipl", "dsl", "civil-code-personal-info"]
domains: ["data-economy"]
original_title: "从“依附”到“独立”：数据资产证券化（ABS）的法律逻辑与进阶展望"
original_author: "武强胜 (零壹法谈)"
original_publication: "零壹法谈 WeChat Official Account"
original_url: "https://mp.weixin.qq.com/s/5-3AdI_W4U45TNaTO6AWgQ"
source_language: "zh"
---
> *Editor's Note — DCC.*
>
> The third and final brief in DCC's data-asset-ABS series turns from what
> these deals *are* to what they are *trying to become*. It draws on the
> forward-looking half of 武强胜's analysis on 零壹法谈. Read it after the
> [mechanism brief](/posts/data-asset-abs-what-is-securitised/) — the "2.0"
> idea only makes sense once you see why today's "1.0" deals run on
> collateral rather than data cash flow.
>
> There is an irony worth stating up front. The [June 2026 exchange halt](/posts/data-asset-abs-suspended-window-guidance/)
> was triggered by deals where the data was a *costume* over conventional
> debt. The fix regulators are signalling — prove the data generates the cash
> flow — is precisely the promise of the 2.0 model below. The trouble is that
> 2.0 remains, for now, largely theoretical. The category is caught between
> the abuse that got it halted and the genuine structure that would justify
> it but does not yet work at scale.

## 1.0 versus 2.0

As the [companion brief](/posts/data-asset-abs-what-is-securitised/) explains,
every issued data-asset ABS is "1.0": the data is **collateral or
credit-enhancement**, and the securities are repaid out of a borrower's
overall operating cash flow via a conventional debt claim. The data's legal
status is *security interest* (担保物).

"2.0" is the version where the data asset becomes the **underlying asset that
directly produces the repayment cash flow** — through outward licensing, data
services, or subscription fees. The shift is fundamental: the data moves from
*collateral* to *base asset*; its value detaches from the holder's credit and
attaches to the **data's own earning power**; and the deal's centre of gravity
moves from *pledge* to *transfer of a revenue right*. Even the most innovative
deal to date — Qingdao's "pure" data-asset ABS — is still 1.0: "pure"
describes the collateral, not the cash-flow source.

## The bridge: secondary licensing (二次许可)

The most plausible route from 1.0 to 2.0 is a structure already proven in
**intellectual-property ABS** — the secondary-licensing model. It runs in
three steps:

1. **First licence — build the base right.** The data holder (the financing
   party) grants an **exclusive licence** of specific data to the originator
   (typically a factoring or finance-lease company). The originator pays the
   **entire licence fee up front** — this lump sum *is* the financing the
   holder receives.
2. **Reverse licence — create the cash flow.** The originator (now licensor)
   grants a **reverse exclusive licence** of the same data back to the
   original holder (now the "client"/licensee), who pays **periodic licence
   fees** into the SPV's monitored account. These recurring fees are the
   ABS's underlying cash flow.
3. **Securitisation.** The originator transfers its licence-fee receivable to
   the SPV, which issues the asset-backed securities; proceeds repay the
   originator's upfront lump-sum outlay.

The result is a closed loop. Economically it resembles a mortgage — the holder
takes a lump sum now and repays in instalments. But two legal features make it
powerful for the data context:

- **The underlying asset is the licence-fee claim, bound to the data's own
  cash flow** — not the holder's general credit. That is exactly the upgrade
  from "enhancement tool" to "base asset."
- **Ownership never moves.** Both the first and the reverse licence transfer
  only *use rights*; the holder keeps title throughout. This sidesteps the
  [true-sale problem](/posts/data-asset-abs-what-is-securitised/) and the tax
  and compliance complications of an ownership transfer — and it neatly avoids
  the very property-characterisation gap that pushes 1.0 deals toward pledge
  structures in the first place.

## Why it does not yet work for data

Porting the IP model onto data runs into four problems — and the first is
close to fatal without careful structuring.

1. **"Exclusive" does not map onto data.** IP secondary licensing depends on
   *exclusivity*: an exclusive patent licence means no one else — including
   the holder — may practise the patent. But data is **non-exclusive and
   infinitely copyable**; the same dataset can be used by many parties at
   once without rivalry. What does an "exclusive licence" of data even mean,
   and can a later exclusive licence reach data the holder already used or
   licensed earlier? The draft [Data Property Rights Registration Work
   Guide](/laws/data-property-rights-registration-guide-draft/) brings a
   "data-operating right" (数据经营权) into the property framework and
   contemplates licensing — an initial institutional basis — but the
   exclusivity question is unresolved at law and untested in court.
2. **Compliance caps what can be licensed.** Data that contains personal
   information, trade secrets, or export-controlled content cannot be licensed
   freely; [PIPL](/laws/pipl/) and the [DSL](/laws/dsl/) constrain the scope,
   term, purpose, and manner. In the *reverse*-licence leg especially, every
   layer must stay inside the **original data subjects' authorisation** — a
   second licence cannot enlarge the consent the first collection obtained.
3. **Licence-fee pricing has no settled basis.** A 2.0 deal needs a defensible
   licence fee, which needs a scientific data-asset valuation. Exchanges are
   still experimenting with data-product pricing; there is no unified standard,
   and (as the [failure cases](/posts/data-asset-abs-suspended-window-guidance/)
   showed) aggressive valuation assumptions are where deals break.
4. **Cash-flow stability is unproven.** Data-licence revenue swings with
   demand, technical obsolescence, and policy — far less stable than
   infrastructure toll income — so it needs heavier structural enhancement
   (over-collateralisation, liquidity-reserve accounts).

## The finance-lease-of-data variant

A second 2.0 path reuses the finance-lease structure — but instead of pledging
data, it tries to make the **data itself the leased object** that throws off
rent, the way a toll road or power station does. One design: a leasing company
buys a de-identified dataset (city-traffic, weather, health) and leases it to a
data-operating company, which earns licence revenue from third parties and pays
rent from that revenue; another is a **sub-lease** (lease in, lease out) that
mirrors secondary licensing with "lease" swapping in for "licence."

The obstacles echo the licensing problems: it is unsettled whether data is an
**eligible lease object** under the Civil Code and finance-lease rules (which
expect a clear-title, real, specific, disposable, income-producing asset);
data's **non-exclusive possession** undercuts the lessor's retained-ownership
premise; and the cash flow is harder to stabilise. Every issued
finance-lease data-ABS so far is, in fact, a 1.0 pledge-enhancement deal; the
genuine lease-of-data structure is still on the drawing board.

## The policy signal

Regulators are, tentatively, pulling in the 2.0 direction. Anhui's
ten-department *Measures to Deepen Data Property-Rights Registration and
Advance the Market-Based Valorisation of Data Elements* (28 March 2026) call
in Article 11 for exploring "**pledge, empowerment, and reverse-licensing**
data-asset securitisation models" — naming reverse licensing explicitly. As
the national [Data Property Rights Registration Work Guide](/laws/data-property-rights-registration-guide-draft/)
moves from draft to practice and rights boundaries sharpen, the
secondary-licensing model could scale. But the
[Data Foundation System Opinions'](/laws/data-foundation-system-opinions/)
three-rights framework still needs substantive-law confirmation before any of
this rests on firm ground.

## What this means for overseas counsel

- **Know where the category is heading.** The destination is *data as cash
  flow*. Today's collateral deals are a transitional form. Multinationals
  building China data-monetisation strategies should design licensing and
  data-operating arrangements now in a way that could later support a
  2.0 securitisation — recurring, metered, contractually documented data
  revenue is the raw material.
- **The non-exclusivity problem is the one to solve in the documents.**
  Because law does not give data real exclusivity, a 2.0 deal has to
  *manufacture* it — through contractual exclusivity, technical access
  control, and monitored delivery — and even then it is a workaround, not a
  settled legal right. Price that residual uncertainty.
- **Licensing compliance is the ceiling.** Only data whose consent scope,
  classification (no important/core data), and — where relevant — cross-border
  posture permit outward licensing can feed a 2.0 deal. The reverse-licence
  leg cannot exceed the original authorisation. Audit the authorisation chain
  before assuming a dataset is licensable.
- **Watch the registration guide and the local pilots.** The national
  registration guide and provincial measures like Anhui's are the leading
  indicators. When the "data-operating right" and its licensing become
  operational and registrable, the 2.0 model gains its missing legal
  foundation — and the [data-pledge-financing](/posts/data-pledge-financing-what-is-pledged/)
  and securitisation landscapes shift together.

## DCC sources

- 武强胜, 《从“依附”到“独立”：数据资产证券化（ABS）的法律逻辑与进阶展望》, 零壹法谈
  ([source](https://mp.weixin.qq.com/s/5-3AdI_W4U45TNaTO6AWgQ)).
- [数据二十条](/laws/data-foundation-system-opinions/) ·
  [Data Property Rights Registration Work Guide (draft)](/laws/data-property-rights-registration-guide-draft/) ·
  [PIPL](/laws/pipl/) · [DSL](/laws/dsl/) ·
  [Civil Code (data & licensing provisions)](/laws/civil-code-personal-info/).
- 安徽省数据资源管理局 等十部门, 《深化数据产权登记 推进数据要素市场化价值化若干举措》
  (28 March 2026), Article 11 (reverse-licensing securitisation models).

> This is an editorial summary of the forward-looking analysis in 武强胜's
> piece, with framing for overseas counsel; the secondary-licensing structure
> and its application to data are the author's. **Not legal advice.**